Abstract: Our main clinical motivation of this project is to provide personalized precision care to patients with jaw (both maxilla and mandible) deformities by significantly improving surgical planning method. The number of patients suffering from jaw deformities is escalating each year. Orthognathic surgery is a main surgical procedure to treat jaw deformities by repositioning bony segments of the jaws. The ultimate outcomes of orthognathic surgery are judged by the final facial appearance. Although the facial soft tissues are not directly operated on, the face changes ?automatically? following the bony changes. Orthognathic surgery requires extensive surgical planning. While we can accurately plan the bony movements and transfer it to the patient during the surgery using computer-aided surgical simulation (CASS) and 3D printing, surgeons are still unable to practically predict the facial changes during the surgical planning, and just hope that a postoperative normal face will be ?automatically? restored. However, this ?mental-clue? approach is not reliable because the facial change does not exactly follow bony change. The problem is even bigger in patients with composite defects. For example, if a patient has a skeletal deformity and mild facial defect, a surgeon must know, before surgery, how to overcorrect the skeleton to camouflage the soft-tissue defect. But this information can only be attained by accurate method to predict facial changes. In addition, from patient?s perspective, the final facial appearance is great concern to them. Therefore, it is extremely important, for both doctors and patients, to accurately predict facial changes. In the previous project period, we have made significant achievements in predicting facial changes following bony movements using finite element (FE) method. However, this approach still requires a considerable amount of time to prepare FE models. In addition, rather than determining the ultimate surgical outcome (the postoperative facial appearance) first, the current method is still to predict the facial change passively following the bony surgery. These hurdles greatly prevent surgeons from practically using it in the clinical setting. Our hypothesis is that a personalized precision treatment outcome can only be achieved if surgeons are able to determine the final treatment outcome, a desired postoperative face, before planning the bony surgery. To test our hypothesis, we propose to integrate outcome-driven and machine learning-based techniques together to first estimate a desired postoperative face, and then plan the bony surgery. The proposed project will have a significant clinical impact on improving patient care quality. It will enable clinicians to develop an optimal surgical plan based on both facial and bony information, on-the-fly, using a single software in their routine clinical practice. It will also revolutionize the surgical planning technique using outcome- driven approach, i.e., to first estimate a desired postoperative face and then plan the bony surgery.